Does Modo Have True Subdivisions

#5

So they are true subdivisions…I take it

Thanx for ur help

Cheers

Alex

#6

You got it.

#7

NURBS are created from Bezier curves which are mathmatically computed. I’m betting that most 3D software use OpenGL evaluators to draw NURBS surfaces. These are not hard polygons so they’ll always look smooth. Sub-D’s on the other hand are hard polygons, when you look at them in flat shading, you’ll see every hard edge. Smooth shading is sort of like an optical illusion. It affects the angle the light hits an edge so that it appears to be smooth.

#8

Um… Silo also has real subdivision surfaces. Dunno where you are getting this “emulated” stuff from.

#9

it sounds like you are confusing true nurb modeler with subdivision modeling. subdivision modeling is true subdivision modeling weather or not it a true nurbs modeler. 2 seperate issues. so if a modeler says sub divisions its true subdivions , i think maya has true nurbs and rhino and electric image modeler .

#10

no. modo doesnt have “true” subdivision surfaces where the subdivided surface is infinitely smooth regardless how close you get to it. modo just has standard subdivisions using the same algorythm to smooth its geometry. the only difference is that its not done to “subdivision surfaces”, but rather standard polygons. so thats why i say that theres no true subdivision SURFACES, but there is true “subdivisions”, because the method they divide by is the same.

as far as i know anyway.

also, last time i checked silo didnt have true subd surfaces either. people often mistake isolines to be something more than it is, but Iso lines are just a simplified way of showing a subdivided model, where the wireframe we see comes from the lowest level on the model. it is still just standard subdivided polygons. so theyre not infinitely smooth surfaces. zoom in and you will see faceting.

#11

ambient-whisper, I can’t help but scratch my head after reading what you said. If you’re speaking in Maya lingo, a “surface” is NURBS. But to the rest of the world when you say “subdivision” we’re thinking hard polygons.

And when you say “zoom in and you will see faceting” that has nothing to do with the way the polygons are divided, that is solely based on how it is shaded. I can take 2 polygons and put them perpendicular to each other and make a 90 degree angle. The crease would be very obvious. But if I tweak the smoothing angle, it would appear to be smoother than a hard crease. The OpenGL shading can also be passed along to the renderer. That is why when you enable smooth shading, and then render a subD mesh, it appears smooth without any faceting.

In every app that I’ve seen advertise sub division modeling tools, it’s hard polygons tesselated X number of times using various algorithms, Catmull-Clarke being one of the popular ones. Lightwave does NOT have sub division modeling. LW uses sub patches. Modo is here to fix that, and give us real sub division.

#12

what most are missing is that just subdividing polygons and using actual surfaces for subdivision are completely different beasts.

ambient-whisper, I can’t help but scratch my head after reading what you said. If you’re speaking in Maya lingo, a “surface” is NURBS. But to the rest of the world when you say “subdivision” we’re thinking hard polygons.
its not just maya lingo.
you can spot true subd surfaces outside of maya too. like in realsoft3d, houdini, renderman, etc.

And when you say “zoom in and you will see faceting” that has nothing to do with the way the polygons are divided, that is solely based on how it is shaded. I can take 2 polygons and put them perpendicular to each other and make a 90 degree angle. The crease would be very obvious. But if I tweak the smoothing angle, it would appear to be smoother than a hard crease. The OpenGL shading can also be passed along to the renderer. That is why when you enable smooth shading, and then render a subD mesh, it appears smooth without any faceting.
its not a shading method. im not talking about faceting because the normals arent smoothed. im talking about the borders of any object. look at the silhouette of any standard subdivided object. you will see the edges around it. it will not be perfectly smooth, no matter how many times you divide the object. zoom in far enough and you will see the difference. thats always been one shortcomming to just subdividing polygons and giving a static result. the data sets get really large, and cant adapt to how far you zoomed in.

take the difference between standard displacement rendering in most renderers…, and the algorythms that renderman uses to render its displacements ontop of true subdivision surfaces.
with true surfaces, zoom in really really close, and the result will be just as smooth as when you were far away.the object gets diced to pixel big polygons and is still rendered extremely fast, with little memory usage.

standard polygon divisions are static, and require you to manually subdivide the object to get it smooth enough for closeups. the downside of this is that your computer might not be able to handle the amount of geometry being thrown around for those closeups. thats why having true subdivision surfaces helps, because they will get diced infinitely, and very efficiently, so you never have to see those ugly edges around the silhouettes of models.

In every app that I’ve seen advertise sub division modeling tools, it’s hard polygons tesselated X number of times using various algorithms, Catmull-Clarke being one of the popular ones. Lightwave does NOT have sub division modeling. LW uses sub patches. Modo is here to fix that, and give us real sub division.

and that works fine for most modelling. giving us an approximization is more than enough while working, but when rendering out for film, standard subdivisions arent enough. the polygon counts of large smoothing iterations can leave little room to do anything else( like give you the ability to store large textures… ) , and they dont render displacements as acurately. thats why having an efficient way to render subdivisions is nice. you can model to your hearts content, the surface will be rendered perfectly smooth, and will still give you tons of ram to work with for high resolution textures, etc.

so anyway. i guess what im saying is that even though modo might not have surfaces that are infinitely smooth, its ok, because we dont need it for modelling work anyway…( unless its really accurate engineering work ) .but it doesnt have the great subd surfaces that can be found elsewhere, which are great for rendering.

as for your modo comment. modo doesnt really do anything new in terms of subdivisions.sure its a step above the limits that LW had, but its nothing really new in terms of modelling, when comparing to other applications.

#13

Ahhh… that certainly clears things up. We just had different expectations.

I just felt that the original poster’s explanation and “expectation” of sub division didn’t do Modo justice. What you’re describing (ambient-whisper) is like the equivalent of vector graphics in 2D. The shape is described in a mathmatical formula to create “infinitely” smooth curves using only a few points. That cuts down on memory usage, and all it is, is a parametric shape rather than a rasterized shape.

At least in my book, sub division is a process of tesselating and approximating polygons. If you want a mesh that’s “infinitely” smooth and behaves like NURBS, guess what, you just use NURBS. It almost sounds like you’re trying to say that “sub division” and NURBS are one in the same, which I do not think is true. And it has been discussed before in other threads that there are no NURBS modeling tools (yet) in Modo.

I hope you understand that I’m not trying to be argumentive, just discussing this topic further. It has the potential of flunking everything I thought I knew about 3D modeling in the past :slight_smile:

#14

I’ll try to wrap this up quick:

The subdivision surfaces are “true” in modo (really, I can’t think of what a “false” subdivision surface would be), as they are in any other package that says they support subdivision surfaces – this describes how the surfaces are generated from the cage, not how they are drawn on the screen. Many renderers tessellate/freeze the implicit model into polygons to render; they do not render the implicit surface itself. This includes OpenGL drawing, which although ti may natively support NURBS (I think it does), it but does not support any form of subdivision surfaces. Beyond that, interactive rendering (like while you are modeling) and final rendering are two different beasts, and thus take different approaches.

There are a myriad of technical details on top of this. Of course, it really all comes down to what it looks like in the final render, and most final renderers dynamically adapt the subdivision level relative to camera proximity for best results.

It should also be pointed out that NURBS can be frozen into polygons as well, and it is quite feasible that some renderers do freeze NURBS into polygons for rendering. Also, modo’s SDS Subdivide option is the same as activating subdivision surfaces and freezing the model – same algorithm, different application.

– Joe

#15

personally. i would like to see something like this become standard in all packages supporting subds. just a little enhancement really, but it makes a huge difference. and it would let us make cleaner cages, without having to distort them anywhere as much to get the smoothed cage to look good. the all too common spiky meshes that are associated especially with beginning artists using subds would dissapear ( hopefully )

http://www.dgp.toronto.edu/people/stam/reality/Research/pdf/eg01.pdf

#16

Of course that’s not true. That’s because SUBD surfaces are more general than NURBS. With NURBS, you’re restricted in topology. This is such an old topic that I shall spend no further discussing this except to say that it’s possible to exactly evaluate the limit surface of a SUBD surface. See the old 1998 paper: http://citeseer.ist.psu.edu/stam98exact.html

#17

Actually, the only renderer that does not approximate SubDs at all is realsoft3d. Even renderman subdivides SDS into micropolygons during rendering.

Cheers,
Mike

#18

There was an interesting thread in our Graphics Programming Forum that discussed the idea of true subdivision surfaces:

http://cgtalk.com/showthread.php?t=118938

#19

“really, I can’t think of what a “false” subdivision surface would be” - quote Jangle-Luxology

I thought that false subdivisions would be like the Proxy Smooth In maya or the cps plugin for maya?

Also arnt sub-patches in modo false subdivs?.. They emulate subdivisions but dont smooth as well or support n-gons

#20

true and false subdivisions…makes it sound like something religious.

no need to make a big deal out of it.

Modo uses the same kind Subdivision surfaces as Lightwave. Which means its just smoothed polygons like Mayas Proxy Mesh.

Subidivision surfaces from an engineering point of view (notice i leave out true/false) are resolution independant just as Ambient Whisper previously posted. And its also possible to model within a heirarchy like in Maya. This is whole other beast mathecially then just hitting tab to get a smoothed polygon.

of course using the LW style of a proxy mesh is enough for a lot of people, its fast and efficient and very simple…but again Ambient mentioned the hangups regarding that.

#21

I think your wrong…Subpatches are the same as lightwaves simulated subdivisions. Like i sed they r proxys…meaning a courser mesh and no support for n-gons

#22

Anyways stuffit (not the app)

Lets just all agree that modo kicks some major ass!

This convo has gone way to far!

Cheers

#23

Even Maya’s suddivs are displayed as polys and you’re working with a cage. If you change selection mode on a subdiv to edge for example, you pick that edge from its control cage and manipulate that.

#24

:slight_smile: anyway. theres no big deal with having real or no subds. it doesnt make your models automatically look better. the main thing that has to look good is the cage. as long as you got a good idea of what the smoothed result will look then youll be fine. theres no reason to see a full perfectly smooth result in the viewport. you only really need infinitely smooth surfaces during rendertime anyway.

( and in most cases you dont even need that :slight_smile: )